Because of the holidays, TI E2E™ design support forum responses will be delayed from Dec. 25 through Jan. 2. Thank you for your patience.

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

DP83848 Auto-MDIX Function Off

Other Parts Discussed in Thread: DP83847, DP83848J

In case of DP83848 Auto-MDIX Function off,

DP83847 data sheet Fifure-8 same connect OK?

  • Are you referring to the Figure 8. 10/100 Mb/s Twisted Pair Interface in the DP83847 datasheet? 

    The correct connections for the DP83848 is shown in Figure 11. 10/100 Mb/s Twisted Pair Interface in the DP83848 datasheet..  

    Patrick

  • Was the information previously provided enough to resolve your questions on the connections?  Is any additional clarification required?

    Patrick

  • Mr.Patrick

    It was settled by your answer. You're very kind. I am sorry that communication was late.

    Uno

  • Patrick-san,

    One of our customers is now facing crosstalk issue on the DP83848 and they doubt the connection between both center tap, Tx and Rx. Therefore they would like to isolate/cut this trace like Fig 8 in the datasheet of the DP83847. Do you think that it is worth trying ?

    Regards,
    Ori

  • No, cutting the connection to the center tap is not recommended.  The DP83848 architecture is different than the DP83847 architecture.  The connections of the terminations shown in the DP83847 datasheet are not correct for the DP83848. 

    What are the characteristics of the crosstalk issue?  How has the customer determined that the problem is crosstalk?  What are they observing in their application?

    Patrick

  • Patrick

    The xustomer issue is as follows.
    When the customer connect Ether HUB to their board where the DP84838 is mounted but power is OFF, the Ether switch devices mounted on the HUB sends Link pulses and wrongly recognizes a reflected pulse via center tap of the transformer as a 10Base-T idle signal, TP_IDL. Then the HUB thinks a 10-Base-T PHY is connected, but it is not actually, then it gets stuck.

    By disconnecting Rx center tap and Tx center tap at the transformer (actually by disconnecting Vdd from Rx center tap), the issue disappears. 

    Here, one question.
    You mentioned that both Rx center tap and Tx center tap have to be connected to Vdd as Figure 7-1 in the At DP83848 datasheet.
    However, the RX path common mode DC voltage is also provided at the point between two 49.9ohms resisters as described in the Figure 7.1.
    Why the Rx center tap has to be also redundantly biased to Vdd?
    What will be wrong if DP84838 is connected as attached schematics?

    Best regards,
    K.Hirano

  • Patrick

    Sorry, I forgot to attache the schematics.

    Best regards,
    K.Hirano

    DP83848.pptx
  • What component is used for the transformer in this application?

    Patrick

  • Patrick,

    I sent the customer’s transformer datasheet to you separately via email.

    Anyway, please let the customer know clear reason why Rx transformer center tap has to be connected to VDD even if DC common mode voltage is already provided through two 49.9ohms resisters.

    Best regards,

    K.Hirano

  • Hirano-san,

    Connecting the transformer center tap to the supply is intended to provide a better common mode point for the transformer.  It is not a strict requirement, but generally provides better performance. 

    Generally, we recommend symmetric transformers and symmetric connections to the transformer.  This way, when the transmit and receive signaling is swapped via Auto-MDIX, the signal paths will still be essentially be the same.

    With that in mind, what is the impact of disconnecting the TX center tap from VDD?  How does this affect the functionality in the customer's system?

    Patrick

  • Patrick,

    Thank you for your response.

    > Connecting the transformer center tap to the supply is intended to provide a better common mode point for the transformer.  It is not a strict requirement, but generally provides better performance. 

    So, this means customer does not need to connect RX center tap of the transformer to VDD (connecting RX center tap to VDD is not mandatory, if center of two 50ohms is connected to VDD near DP83848), if desired. Is this understanding correct?

    > With that in mind, what is the impact of disconnecting the TX center tap from VDD?  How does this affect the functionality in the customer's system?

    The customer does not want to disconnect TX center tap from VDD at all.
    The customer wants to disconnect RX cent VDD to prevent a HUB malfunction due to reflection via RX/TX center tap connection.

    Best regards,

    K.Hirano

  • Hirano-san,

    I inquired about the TX center tap because I am trying to improve my understanding of the functionality of this system and, specifically, the symmetry of the transmit and receive paths.  Normally, we try to make sure that these paths are very symmetric. 

    Could you confirm that Auto-MDIX will be disabled for this application and that the TX path will only be used for transmit?  I realize that the title of this post includes the phrase "Auto-MDIX Function Off", but I wanted to confirm this is the mode in which the application will operate.  If so, then my question about the TX center tap may be irrelevant. 

    Your understanding regarding the RX center tap is correct.  The DC common mode of the RX center tap will be provided via through the connection of the 49.9 Ohm resistors to VDD.

    Patrick

  • Patrick,

    I confirmed that the customer is using DP83848J and MDIX_EN (#34pin) is externally pulled down, it means Auto-MDIX mode is disabled.
    So, the customer does not need to connect transformer RX center tap to VDD, right?

    Anyway, one more question.
    The TX DC common mode voltage is provided via center of two 49.9ohms too.
    What if both TX center tap and RX center tap of the transformer are disconnected from VDD?
    Connecting transformer TX/RX center tap to VDD looks redundant.
    What would be wrong if they are disconnected from VDD actually?

    Best regards,

    K.Hirano